Article pubs.acs.org/est
Framework for Resilience in Material Supply Chains, With a Case Study from the 2010 Rare Earth Crisis Benjamin Sprecher,*,†,‡ Ichiro Daigo,§ Shinsuke Murakami,⊥ Rene Kleijn,‡ Matthijs Vos,∥ and Gert Jan Kramer‡ †
Materials Innovation Institute (M2i), Delft 2600 GA, The Netherlands Institute for Environmental Sciences (CML,) Leiden University, Leiden 2311 EZ, The Netherlands § Department of Materials Engineering, The University of Tokyo, Tokyo 113-8654, Japan ⊥ Department of Systems Innovation, The University of Tokyo, Tokyo 113-8654, Japan ∥ Faculty of Biology and Biotechnology, Ruhr-Universität Bochum, Bochum 44801, Germany ‡
S Supporting Information *
ABSTRACT: In 2010, Chinese export restrictions caused the price of the rare earth element neodymium to increase by a factor of 10, only to return to almost normal levels in the following months. This despite the fact that the restrictions were not lifted. The significant price peak shows that this material supply chain was only weakly resistant to a major supply disruption. However, the fact that prices rapidly returned to lower levels implies a certain resilience. With the help of a novel approach, based on resilience theory combined with a material flow analysis (MFA) based representation of the neodymium magnet (NdFeB) supply chain, we show that supply chain resilience is composed of various mechanisms, including (a) resistance, (b) rapidity, and (c) flexibility, that originate from different parts of the supply chain. We make recommendations to improve the capacity of the NdFeB system to deal with future disruptions and discuss potential generalities for the resilience of other material supply chains.
1. INTRODUCTION Material resource constraints are an important aspect of sustainability. The literature often highlights rare earth elements (REEs) because of their importance to the high-tech industry, the fragility of their supply chains, and environmental concerns.1−3 Of particular concern are neodymium and dysprosium, used predominantly in high performance neodymium−iron−boron (NdFeB) magnets.4 This relatively tiny sector found itself becoming a key supplier to the high-tech industry. With neither the capital nor the expertise to properly respond to the tech sector’s booming demand, the NdFeB supply chain was already strained when a sharp reduction of Chinese rare earth export quotas and a short-term boycott caused major upheaval among REE end-users.1 It is not surprising that the NdFeB industry bore the brunt of this 2010 REE crisis, with prices flying wildly out of control, in some cases increasing by an order of magnitude in the span of a few months.5 The crisis provoked a multitude of reactions across the entire NdFeB supply chain, ranging from dozens of junior mining companies claiming imminent rare earth production to end users reducing their reliance on neodymium magnets, or even substituting NdFeB completely.6,7 The sum of these events resulted in prices falling significantly, if not actually reaching © 2015 American Chemical Society
pre-2010 levels. Although the REE sector has many idiosyncrasies, when looked at from afar this type of boombust dynamics can often be observed when small raw material supply chains are integrated as supplier into a major industry.5 As the dust of the 2010 REE crisis was settling numerous reports and scientific publications investigated the rare earth sector, sometimes with diametrically opposed conclusions. For instance, Gholz (2013) writes that “the largely successful market response” offers the lesson that “policymakers should not succumb to pressure to act too quickly or too expansively in the face of raw material threats.”8 However, Tukker (2014) concludes that “Western governments ignored market failures” resulting in the fact that the Western world was “entirely outmaneuvered by an economy that was guided.”1 More cautious analyses are made by Machacek and Fold (2014),9 who focused on the efforts of Molycorp, Lynas, and Great Western Minerals group to build a Western primary REE supply chain. They give a good historical overview of events and conclude that the bottleneck for the establishment of Received: Revised: Accepted: Published: 6740
January 13, 2015 May 11, 2015 May 12, 2015 May 12, 2015 DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
system in terms of resilience, and shows how and where distinct resilience concepts apply in the various stages of the NdFeB supply chain. A follow up paper will focus on providing indicators and will quantitatively assess the resilience of the NdFeB system during the 2010 REE crisis.
alternative REE supply chains is at the chemical separation phase, because this technically challenging process is both expensive to build and operate. Golev et al. (2014)10 contribute a broader overview of several non-Chinese supply chains, discussing their opportunities and constraints in increasing primary production while also noting that there is industrial interest in recycling. Both Rademakers (2013)11 and Sprecher (2014)12 investigated the recycling opportunities for NdFeB, showing that recycling of magnets will most likely have limited effect on global supply. Seo and Morimoto (2014)4 compared recycling and substitution strategies from a Japanese perspective and concluded that substitution makes more sense because most of the Japanese REE demand is subsequently exported and not available for recycling to the Japanese industry. From this body of literature it is clear that the REE supply chain is a complex and intricately interlinked system. However, previous work has for the most part only analyzed small parts of the system in relative isolation. We are not aware of any work that takes a broader systems perspective and uses a rigorous theoretical framework to analyze the resilience of the NdFeB system as a whole. This perspective allows the analysis of different aspects in a wider system context, while also investigating interactions between different parts of the supply chain. We feel this is especially relevant because it provides an interesting case study for more generally applicable insight into the ways supply chains of critical materials can respond to resource constraints and disruptions. On the basis of an extensive literature review and interviews with actors ranging from large players such as Siemens and Hitachi Metals to individual entrepreneurs and government agencies, we developed a framework that aims to improve our understanding of how material supply chains respond to supply constraints and disruptions. We were guided by two research questions:
2. METHODS Using a semistructured interview format, we first interviewed actors from across the NdFeB supply chain, as well as governmental and academic experts. Fifteen interviews were conducted in total. Seven interviewees were from Japan, seven from the E.U., and one from the U.S. We then used the input from the interviews to build a qualitative representation of the NdFeB supply chain, using the system dynamics methodology introduced by J. Forrester.16 As far as possible, we used the same terminology as commonly used in material flow analysis.17 This allowed us to map the socio-economical drivers on the physical flows in the supply chain, and visually represent at what places in the physical supply chain different mechanisms contribute to resilience. NdFeB magnets often contain other rare earths in addition to neodymium, most notably praseodymium and dysprosium. For the sake of readability, we will address these collectively by using the terms NdFeB and neodymium, except when relevant (in section 4.3). Our results are based on information provided by the interviews, except where references to specific sources are given. The developed framework is our own work, that served to interpret the provided content in view of resilience theory. The full list of interviewees and the semistructured interview questionnaire can be found in the Supporting Information (SI). 3. RESILIENCE AND THE NdFeB SUPPLY CHAIN In section 3.1 we define supply chain resilience. Section 3.2 introduces rapidity, resistance, and flexibility as the three system traits that together give rise to supply chain resilience. In section 4, we discuss the concrete mechanisms found in the rare earth supply chain that underpin rapidity, resistance, and flexibility. 3.1. Definition and Features of Supply Chain Resilience. A more resilient system has properties that allow it to show limited consequences from disruptions and fast recovery times.18 In the context of material supply chains, we define resilience as the capacity to supply enough of a given material to satisf y the demands of society, and to provide suitable alternatives if insuf f icient supply is available. In practice, this means that both the supply and demand of NdFeB will need to have a certain elasticity, which allows the system to absorb supply or demand disruptions without significant price fluctuations. Although the literature mentions reduction of failure probabilities as playing an important role in enhancing resilience,19 we stress that in the context of this framework we see resilience as the sum of several generic system dynamics, observable in material supply chains. These dynamics together enhance the overall response of the system to any kind of disruption, whether foreseen or unforeseen. Our work could suggest that resilience is always good thing. However, whether this is true depends both on the boundaries of the system under investigation and the time scale under analysis.13 It is worthwhile to note that the flip side of resilience
(1) What type of mechanisms along the NdFeB supply chain provide resilience in response to supply constraints and disruptions? (2) What system perspective based policy recommendations can be made to improve the capacity of the NdFeB system to deal with future constraints and disruptions? In our research, we used resilience theory as a framework to interpret the information we gathered from the interviews. Resilience can be defined as the capacity of a system to tolerate disruptions while retaining its structure and function.13 Further below, we develop a framework for the interrelated and complementary mechanisms that provide resilience in material supply chain, when these are confronted with resource constraints and disruptions. In the remainder of this work, we use “disruption” to refer to quick, short-term supply disturbances and “constraint” to refer to slower, long-term disturbances. The resilience of systems has long been a subject of research, and is a recognized feature of many types of systems ranging from ecological to socio-technological.14,15 We hope to advance resilience research by developing a novel and concrete framework that defines and clarifies resilience in material supply chains: novel, because we are not aware of any research on the combination of resilience theory and resource constraints; and concrete, because we made extensive use of knowledge of the physical flows in the NdFeB supply chain. Our framework conceptualizes the dynamics of the NdFeB 6741
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
Figure 1. Resistance and rapidity depend on the strength of the system against disturbances. In this case, the NdFeB supply chain, highlighted in red. Flexibility is the ability to switch between subsystems, and can occur on all levels of the overall system, highlighted in orange. Examples of alternative energy related supply chains are outlined in blue.
between alternative systems that can provide the same service. More formally these are defined as follows: • Resistance; the system maintains it function, i.e., it is able to tolerate various types of disturbances without experiencing unacceptable loss of function. • Rapidity; the system is able to rapidly recover so that it meets its goals again within a short period following the disturbance. • Flexibility; the system is capable of meeting supply needs under a disturbance by switching between different (alternative) subsystems. The supply chain resilience framework is a clear example of industrial ecology, as all of the three resilience-contributing factors above have direct counterparts in ecology: “resistance” is used to describe how ecological systems remain “essentially unchanged” under disturbance; “rapidity” is often defined in terms of ‘return times to equilibrium’ or in terms of the closely related resilience measure “1/return time”, and “flexibility” is used as such to refer to how consumers in ecological food webs switch between alternative resource types.15,21−24 Let us consider the previous example of energy: given the fact that society will need to switch to a sustainable source of energy to avert catastrophic climate change, it is desirable to have a resilient sustainable energy sector capable of meeting the rapidly increasing demands of society. Wind turbines are one of the main options for producing sustainable energy. Modern wind turbines can use either geared or direct drive technology. The latter utilizes a large amount of NdFeB magnets, while the former requires specialty metal alloys for the gearboxes. Direct drive wind turbines are expected to increase in market share because they allow for higher efficiency and lower maintenance costs.25,26 We would consider the wind turbine industry resilient if it is capable of providing sufficient wind turbines to fulfill the needs of society, even in the face of exponentially increasing demand and potential constraints and/or disruptions. The resilience of the wind turbine system depends on the ability of the NdFeB system to provide a sufficient quantity of magnets to fulfill the demand for direct drive wind turbines (resistance) in the face of disruptions and/or constraints, and, if
is lock-in. Thus, short-term desirability for resilience can come in conflict with long-term desire for system change. For example, the energy system could become more resilient, if more unconventional fossil sources were put into production. In the short term and from a local perspective, the improved stability of energy infrastructure (energy security) is a positive. For instance, the increased production of light tight oil in the US has weakened the OPEC monopoly. However, if one enlarges the system boundaries to include the long-term perspective of society as a whole, the benefits of these huge investments in fossil energy and related infrastructures are less clear, because the lock-in created by these investments makes it more difficult to move away from using fossil fuels, thereby making our society less resilient against climate change. Here we note that the ecological literature offers a way out of lock-ins through adaptive capacity on a longer time scale (provided by a purging of nonfunctional system traits and innovation introducing new functional system traits), as it provides novelty to the functioning of supply chains. However, this mechanism was not mentioned by any of the experts interviewed. 3.2. Resistance, Rapidity, and Flexibility: The Cornerstones of Resilience. In this section, we introduce the concepts of resistance, rapidity, and flexibility,19 and use these to discuss how our case study relates to the larger socioeconomical system it is embedded in.20 But first we need to define our system boundaries clearly. We conceptualize our system as having three levels: • Society; which has certain needs, such as transportation or energy. • The production system; the system that meets the needs of society and is responsible for converting materials into services. For example, the need for sustainable energy can be met by the producers of wind turbines. • The NdFeB supply chain; the system that provides the materials required by the production system to provide wind turbines to society. We conceptualize the resilience of a system as depending on factors that either allow it to directly maintain function under disturbance, to rapidly recover from a disruption, or to switch 6742
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
Figure 2. NdFeB supply chain. Each box represents a stock of neodymium, while the form in which the neodymium is in that part of the supply chain is indicated between brackets. The arrows between the stocks represent physical flows of material, and are named according to the process used to transform the neodymium to its next form. Green indicates stocks and flows that do not currently happen on a relevant scale (“postconsumer recycling”).
Figure 3. Diversity of supply: red indicates a system disruption or constraint. The blue arrows represent how the elements in the system influence each other: the arrowheads indicate the direction of the influence, the S or O next to the arrowhead indicate whether the connected parameters change in the Same or Opposite direction.
As a reference point, Figure 2 shows the physical supply chain in the visual language of mass flow analysis. A more detailed description of the supply chain can be found in Sprecher et al. (2014).28 4.1. Diversity of supply. The Chinese export quotas were especially problematic because at that point in time China controlled 96% of the world REE production. Within China, two-thirds of rare earth oxides production originates from the Bayan Obo mine in Inner-Mongolia.29 Clearly, such a narrow supply base is not robust. Therefore, we introduce “diversity of supply” as the first feature of a resilient material supply chain (note that this feature is closely linked to the concept of redundancy in resilience literature,30 and to switching capacity. In this case not between alternative supply chains, but alternative providers of the same raw material). In terms of system dynamics, having high diversity reduces the impact of a given disruption of the supply of REE ore to the rest of the supply chain. Figure 3 shows how we integrate diversity of supply into the NdFeB supply chain, which we conceptualize as the sum of primary production, postconsumer recycling and smuggling. Note that diversity of primary supply is only useful if the subsequent actors in the supply chainin this case the REE refinerieshave the technical and organisational capacity to switch timely between suppliers. Primary Production. The most obvious way to increase diversity is to have more mines, and to build mines in different countries. However, as Chinese attempts to buy REE mines in Australia and Greenland show, one should also pay attention to ownership issues, not just location.
the NdFeB supply chain fails, on the speed with which the supply chain can recover (rapidity), or on the ability of the system to switch from producing direct drive to geared wind turbines (flexibility). Figure 1 shows how resilience of the sustainable energy system depends on the resistance and rapidity of the actors within each level of the larger socio-economic system, as well as on flexibility between these levels. If demand for sustainable energy grows so fast that both the direct drive and the gearbox supply chains are not capable of keeping up with demand, then society has the choice to use an alternative source of sustainable energy. For example photovoltaic energy, which will invariably have its own supply chain challenges. A real-world example is Siemens, whose current generation of wind turbines is of the direct-drive type. It has invested in takeoff agreements with the rare earth industry to ensure access to neodymium.27 However, it also has geared wind turbine designs, ensuring that it can switch between alternative supply chains if this becomes necessary. Competing wind turbine producer Enercon has invested in direct-drive technology that functions without NdFeB magnets, using a synchronous generator and an electrical rotor instead.25
4. MECHANISMS OF RESILIENCE IN THE NdFeB SUPPLY CHAIN In the previous section, we discussed at an abstract level how resistance, rapidity, and flexibility contribute to resilience in material supply chains. In this section, we use system dynamics to discuss which concrete resilience mechanisms we identified in the NdFeB supply chain. 6743
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
Figure 4. Supply/demand mechanism: feedback loops are represented with a circular arrow. The number is used to identify the feedback loop for further discussion. Feedback loops are identified by their number (B1), where the B indicates that this is a balancing feedback loop. The double dashed blue arrows indicate that there is a delay in the influence.
Recycling. We distinguish between two types of recycling: pre- and postconsumer recycling. In Figure 3, the green arrow represents the (at the moment of writing) mostly hypothetical option of recycling postconsumer waste magnets. Because of quality concerns, postconsumer recycling would in all probability lead back to the REE refinery stage, where the rare earths are extracted via acid leaching.28 Postconsumer recycling increases the diversity of supply because it can complement primary production. In contrast, preproduction recycling (the black arrow in Figure 3) of material lost during manufacture of the magnets (e.g., grinding losses) should be seen as a measure to increase production efficiency, which does not affect diversity of production. Illegal Mining and Smuggling. Besides the legal export of Chinese REEs, illegal sources can also provide a significant supply of raw material, estimated at one point to add 40% to official Chinese production.31 This illegal material is either used in China or exported in the form of ore concentrates or oxides. The option of acquiring illegally exported material can reduce the impact of a supply disruption, but we note that smuggling and illegal mining go hand in hand with enormous social and environmental problems.32 To simplify the figure, we assumed all smuggling was in the form of concentrates. In India, illegal export of REE bearing monazite ore has also been reported in press, albeit at a smaller scale.33 Other Options. Finally, we would be remiss not to mention more exotic options such as deep-sea mining,34 which, despite being challenging for technical, regulatory and environmental reasons, have significant disruptive potential because of the enormous amount of metals they could release into the market. Because interviewees indicated this is most likely still decades away we did not include it in our formal framework (interview Dr. Jiro Yamatomi). 4.2. Feedback Loops through Price Mechanism. There are a number of feedback loops throughout the NdFeB supply
chain. Here we will discuss the main feedback loop: the supply/ demand price mechanism, shown in Figure 4. Although in reality each actor in the supply chain has an associated supply and demand, we simplify this by representing supply as “ore supply” at “primary production”. This simplification still captures the essential system dynamics because, based on interviews, we infer that supply side constraints mostly exist at the beginning of the supply chain. Increasing “material demand” leads to a higher “neodymium price” and vice versa, while a high price will depress demand. Similarly, less availability of “ore supply” causes an increase at “Neodymium price”. This in turn will lead to “investments in new primary production”, which, after a significant time delay, increases the ore supply. Because it is not possible to process a single REE without processing the majority of the associated REEs,35 the primary production feedback-loop is complicated by the demand for other REEs. These are part of an identical feedback loop that also influences the decision to invest in new primary production. Although an important feature of many minor metals, the issue of comining is vital in rare earth economics. This means that increasing production of one high-demand REE will lead to overproduction of the associated REEs. “Other REE demand” is colored red, to indicate that it falls outside the system boundaries of this model. Note that in the full system dynamics picture, neodymium price also drives other factors such as investments in recycling. However, for clarity’s sake we only describe the main feedback loop here. There are numerous other feedback loops discussed in the remainder of this work. 4.3. Material Substitution and Improved Material Properties. On the material demand side, a number of options are discussed in the material efficiency literature, such as increasing the lifetime of products, increase the use intensity of products (e.g., through products service systems) and the reuse of components.36 The NdFeB supply showed a more limited response with respect to reducing material demand. Interviewees identified 6744
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
Figure 5. Material properties and substitution: these have the same function (creating balancing feedback loop with price) but are implemented by different actors.
Figure 6. Stockpiling mechanic: introducing the parameter “perceived short-term threat of supply disruption”. This type of parameter is distinct from others in the model, because it shows how beliefs of actors in the system influence behavior. It also features a reinforcing feedback loop (R1).
the level of product design and relates to flexibility, while improving the material properties is done at the level of the magnet producers and relates to resistance. Figure 5 also shows how demand changes are incorporated into the model. Societal trends, such as increased demand for sustainable energy or smartphones, may change over time, leading to demand changes. At a lower system level, component changes may also lead to changes in demand (such as replacing LCD with OLED screens). There are also two balancing feedback loops: substitution and improved material properties both reduce the demand for neodymium and/or dysprosium. This causes the price of NdFeB to go down, which in turn will lessen the need for further substitution or property improvement. Conversely, a low REE price can also lead to inefficient material use and cheaper production techniques that yield lesser material properties. 4.4. Stockpiling. Stockpiling can improve the resistance of the system, because a stockpile can absorb sudden price and/or supply fluctuations. However, stockpiling can also have a detrimental effect. During the 2010 REE crisis some Japanese companies forced their suppliers to increase their stockpile of raw materials to up to two years, at the very moment the prices were highest and the materials were hardest to obtain. This drove the price of neodymium and dysprosium up significantly (interview Hitachi Metals). Because there are many different grades of NdFeB magnets, each with slightly differing alloying element ratios, it is difficult
two mechanisms that actors used to change their neodymium consumption. Substitution. Substitution is the well-known switching mechanism whereby one material is substituted for a different material. There are many levels where substitution can occur, ranging from using a lower grade of the same material to outright substitution of the entire technological system dependent on that material (e.g., replace wind energy with PV). On the basis of our interviews we highlight the two most common types of substitution: • Material substitution: the case where the requirement of using magnets remains in the final product design, but this requirement is fulfilled with a different material (e.g., replacing NdFeB magnets with samarium−cobalt magnets). • Technological substitution: where a product is redesigned to operate without any magnets at all (e.g., replacing a direct drive with a geared wind turbine). Changing Material Properties. Improvement of the properties of materials with the goal of reducing material usage represents a less drastic but more often realized measure (e.g., using grain boundary diffusion technology to allow magnets with lower levels of dysprosium to have equivalent high temperature operational specification).4 The distinction between substitution and material properties is relevant because they are different types of actions taken by different actors. As shown in Figure 5, substitution is done at 6745
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology Table 1. Types of System Disturbances fast slow
supply
demand
natural disaster, political issues protective measures, ore depletion
disruptive demand change societal, technological trends
Russia releasing their inventories, which caused an oversupply situation.5 A more long-term oversupply situation can currently be observed in the cerium market, an REE that is coproduced with neodymium. The increased demand for neodymium resulted in a cerium glut, severely depressing prices (interview Nissan). On the demand side, relatively slow constraints come from societal and technological trends, such as increasing electrification and use of sustainable energy. Fast demand increases usually stem from either an exploding demand for a new type of product (e.g., smartphones), or from component changes in existing products (e.g., new generation of wind turbines switching from geared to NdFeB containing direct drive technology). 5.2. Options to Improve Supply Chain Resilience. Since the 2010 crisis, several alternatives to Chinese primary REE production have come online, indicating that this is a mechanism that the NdFeB supply chain naturally resorts to in order to solve constraints. In this section, we describe various other policy options to improve the resilience of the NdFeB supply chain. Reduce Red Tape To Improve System Response Times. Although laws and regulations often exist for a good reason (e.g., ensuring that the social and environmental consequences of new economic activities are well understood) they can also act as an impediment to system change. The long lead-time in obtaining permits is a frequently mentioned example of red tape leading to delay in building new primary production. On the recycling side, an example of regulatory issues are the European waste laws. These can obstruct recycling because once a material is labeled as a waste it is difficult to use it as an end product again. Companies and governments should work together to reduce the impact of regulations on the time needed to implement solutions, for instance through doing some of the work already in advance. For example, the German Rohstoffallianz provides its members with options to bundle interests from a value chain perspective and optimizing the supply planning horizon, for instance by drafting templates for framework takeoff agreements (interview Rohstoffallianz). Implement a Mineral Tax. One of the challenges with implementing a recycling scheme is that the prices of raw materials are often too low to warrant recycling. The production costs of primary production are lower than the collection and processing costs of recycling. Because of the supply/demand feedback loop, primary production will outcompete recycling. This is of course assuming sufficient reserves for increasing primary production, which for REEs certainly is the case. The same effect can be seen at the product design level with regard to efficient use of materials. The principal reason for a mineral taxation scheme levied at primary production is to redistribute the profit made from exploitation of nonrenewable resources.40 However, if a mineral tax were to be implemented on a significant (if not global) level, a secondary effect would be that the increased costs of primary production can prevent the supply/demand feedback loop from steering the NdFeB system away from more
for magnet producers to keep significant stockpiles (interview Arnold Magnetics). Stockpiles are usually kept in the form of REE oxides, by the companies producing alloys. Stockpiles also exist at country level. For example, the 2013 biannual US Strategic and Critical Materials report recommended to stockpile $120 million in heavy rare earths,37 and the Japanese independent administrative institution JOGMEC holds a 42 day stockpile for nine metals (Ni, Mn, Cr, Mo, W, Co, V, In, and Ga, but not REEs). Japanese companies are obliged by law to hold an 18 day stockpile.38 Finally, stockpiling can also be employed by speculators who aim to benefit from price volatility. In our resilience framework, we represent the stockpiling dynamic by adding a physical stockpile at the level of the REE smelter and a “perceived short-term threat of supply disruption” parameter. Speculation played a (limited) role in driving up prices during the 2010 crisis39 and is represented here at the same level as emergency stockpiling. Representing both the positive and negative effects of stockpiling, there are two competing feedback loops governing this mechanic: • Reinforcing feedback loop: a supply disruption and/or a sharp increase in price increases the “perceived shortterm threat of supply disruption”, which leads to emergency stockpiling by manufacturers and speculation. This drives up the material demand, which in turn increases price. A strong price increase in itself will fuel the perceived threat of supply disruption, leading to more emergency stockpiling. • Balancing feedback loop: physical stockpiles act to reduce “perceived short-term threat of supply disruption”. Increasing this stockpile through emergency stockpiling will reduce the perceived threat of supply disruption, causing a reduction in the need for emergency stockpiling.
5. NDFEB SUPPLY CHAIN SYSTEM DYNAMICS In this section, we look at the NdFeB supply chain from a complete systems perspective. First, we discuss the different types of system constraints and disruptions, then the various system level interventions that could be implemented to improve resilience. Finally, we combine the resilience mechanisms into a single system dynamics representation of the NdFeB system that shows how the various elements interact with each other. 5.1. Types of System Constraints and Disruptions. On an abstract level, there are two types of system disturbances: those that affect supply and those that affect demand. These disturbances can range from fast to slow. As shown in Table 1, a sudden disruption of supply could be the result of natural disasters, such as the 2011 flooding in Thailand, or political issues such as the Chinese rare earth embargo of 2010. In the long term, supply constraints could be caused by ore depletion or policies like export quotas and taxes. A sudden increase in supply can come from governments releasing stockpiles. This happened for instance with tungsten in 1995: a sharp increase in price led to China, Kazakhstan, and 6746
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
Figure 7. System dynamics of the NdFeB supply chain: this figure combines the resilience mechanisms from the previous section and adds some elements, as discussed in the text. See supporting information for a larger version of this figure.
the overall value of the products they are contained in, meaning that a stockpile could acts as a relatively cheap insurance policy.1 However, as described in section 4.4, it does not make sense for end-users to stockpile neodymium, but rather the alloy producers, for whom the cost of neodymium is a very significant barrier to stockpiling. Although there is some adhoc stockpiling based on individual agreements between endusers and magnet producers (interview Arnold Magnetics), we suggest a common stockpile would be an efficient way to solve this problem. This stockpile can come with a prearranged protocol on how to divide its contents in case of an emergency. This would help to prevent actors from driving up the price by chasing the same stockpile, as happened in 2010.. 5.3. Complete System Dynamics of the NdFeB System. In Figure 7, we combine the previously introduced system elements into a complete overview, which shows how and where the various resilience mechanisms interact with the physical stocks and flows of the NdFeB system. The green elements show the options to improve resilience, while the red elements show the various types of disturbances. Both disturbances and mechanisms to deal with disturbances are distributed across the entire system. Having all of the various elements of resilience together in one place illustrates that every part of the supply chain is somehow involved. Insofar as that, especially in the aftermath of the 2010 REE crisis, the individual actors have relatively little information on the behavior of other actors in the system, we consider resilience to be an emergent system property. Compared to the individual resilience mechanisms, there are also some minor additions. In order for recycling to the “Neodymium price” to work, we add the “investment in recycling infrastructure” parameter that mirrors the “investment in new primary production” parameter.
sustainable material use, especially if some of the tax revenue would be used to support recycling. Support R&D to Expand Use of Excess REEs. A low consumption of other REEs can prevent investment in primary production. This is also known as market inelasticity, where an increase in demand does not translate automatically to an increase in supply. In order to solve this balance problem, one could stimulate the demand for other REEs through focused R&D to find new applications. For example, there is a trend to alloy magnesium with REEs to improve the creep resistance of magnesium alloys.41 Creep resistant magnesium alloys are used for drive train applications in the automotive industry for purposes of weight-saving.42 Promote Design for Recycling. Products using NdFeB magnets are currently designed in such a way that separating the magnets is very difficult.12 This is reflected in the fact that even when the price of neodymium increased dramatically, postconsumer recycling did not take off in any meaningful way. Even if recycling of some applications with large volumes of magnets would become feasible (e.g., wind turbines, electric vehicles), this still leaves the many applications of smaller magnetsweighing no more than a few gramswhere recycling would be uneconomical under almost any circumstance. This also applies for the many other critical elements that are used in very low concentrations. Implementing design for recycling principles and having waste regulations that are not mass based targets, but regulate which materials need to be recycled would in all probability improve this situation (interview Dr. Allan Walton). Increase Stockpiling. Stockpiles offer the possibility to completely negate the impact of any temporary supply disruption, albeit at significant capital costs. Especially for neodymium (and rare earths in general) there is a case to be made for stockpiling, because their costs are but a fraction of 6747
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
The main stabilizing/destabilizing forces in the system are the feedback loops, of which the economic feedback loop (i.e., price mechanism) is the most important. Figure 7 gives an overview of how all the feedback loops and mechanisms are connected to the supply chain. Not all responses to the 2010 REE crisis contributed positively to system resilience. We note the two most explicitly negative responses. The first is panic buying by Japanese companies, who tried to increase their stockpile only after the Chinese export quotas came in full force. This contributed greatly to the price increases. The second is illegal mining and smuggling of Chinese rare earths (estimated at 40% of the official production).31 Although smuggling increases the diversity of supply and thus increases the resilience of the sector, illegal mining has devastating environmental and social effects.32 In the past several years, the diversity of primary production has improved significantly, with several new primary production sources of REE becoming operational.9 However, increasing the diversity is only one, potentially limited and exclusively supply side focused lever that can be pulled in order to improve the capacity of the NdFeB supply chain to deal with future constraints and disruptions. We proposed five additional system interventions: • Reduce red tape for faster system response times (i.e., legislation related to mining permits, recycling) • Implement a mineral tax to promote more sustainable use of raw materials • Support R&D to expanded use of REEs that are comined in excess • Promote design for recycling • Increase stockpiling to effective levels Improving the rapidity of the system can be achieved by improving the robustness of production facilities against natural/operational disasters. This did not come up in any of the interviews, probably due to the fact that we did not succeed in interviewing actors related to the first stages of the supply chain, predominantly located in China. In section 1, we discussed how Gholz (2013) argued that the NdFeB case study shows a “largely successful market response”8 while Tukker (2014) wrote that it shows how “western governments ignored market failures”.1 Out of the above five system interventionsbased on suggestions from actors in the NdFeB supply chainonly “reduce red tape” is in favor of further improving the free market response, while the other four relate to intervening in the free market. From this, we tend to agree with Tukker, that our case study indeed contains a certain amount of market failure. Finally, we would be remiss not to discuss our framework in relation to the various critical materials methodologies that have recently been proposed. Most notably by Graedel et al. (2012), who present a very thorough analysis of how to measure and rank the criticality of metals, applied to REEs by Nassar et al. (2015).44,45 Although there are dissimilarities in time frame and system boundaries, their dimensions “supply risk” and “vulnerability to supply disruptions” strongly overlap with our supply and demand side resilience mechanisms. However, there are significant conceptual differences. Our work is focused on the dynamic aspects of the supply chain; how it changes over time in response to disturbances, while the Graedel et al. framework essentially generates a static snapshot of criticality. The latter
Investments in primary production or recycling are long-term projects, depending not only on the spot price of rare earths, but also on the expected long-term demand. We model this by making long-term investments dependent on the “perceived long-term supply risks” parameter, which accounts for slow trends such as depletion of existing mines, societal trends, technological developments, and protective measures. One might expect that long-term supply risk is something that influences material selection choice at the product design stage, but according to our interviews this is not the case, and materials are selected solely on their economic and physical properties (interview Chatham House). Finally, there is a minor feedback loop connected to “perceived long-term supply risks” that covers the legal responses to protective measures by states, mainly in the form of WTO lawsuits.
6. DISCUSSION In this paper, we provide a framework that defines and clarifies supply chain resilience. We demonstrated the use of this framework by analyzing the multiple responses of the neodymium magnet (NdFeB) supply chain to the 2010 Chinese export restrictions. As a consequence of these restrictions, the price of neodymium increased by a factor of 10, only to return to almost normal levels in the following months, despite the fact that the restrictions were not lifted (export quota have been lifted since 1 January 2015, but export licensing may substitute its restrictive effect).43 These events indicate that the NdFeB supply chain was not very resistant to the disruption, but it recovered with remarkable rapidity. Compared to previous literature discussing REE supply chains, our framework allows a more nuanced and complete analysis of how supply chains of critical materials respond to disturbances. Besides contributing to the understanding of NdFeB supply chain system dynamics, we also believe the framework to be of more generic relevance to those interested in material criticality and the resilience of material supply chains. Through literature review and extensive interviews with actors across the NdFeB supply chain, we have shown that resilience in the NdFeB supply chain is composed of resistance (the ability to tolerate disturbances without unacceptable loss of function), rapidity (the ability to rapidly recover from a disruption), and flexibility (the ability to switch between alternative subsystems). We found that the following concrete mechanisms are primarily responsible for this resilience. On the supply side: • Diversity of supply; more variety in sources of raw material potentially reduces the impact of a disruption or constraint on the remainder of the supply chain. • Stockpiling; acts as a buffer that lessens the impact of temporary supply disruptions. On the demand side: • Improving the material properties; magnet producers have responded to supply constraints by improving the properties of NdFeB, greatly reducing the required amount of dysprosium for high temperature resistant magnets. • Substitution; some producers substituted NdFeB magnets with other magnets, while others switched to a completely different technology that did not rely on permanent magnets. 6748
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology
(5) Metal Prices in the United States Through 2010; Scientific Investigations Report 2012−5188; U.S. Geological Survey, 2013. (6) Hatch, G. P. The Impending Shakeout In The Rare-Eart Sector: Who Will Survive? Technology Metals Research, 2013. (7) Harman Continues Quest For Neodymium Substitutes in Speaker Magnets. Consumer Electronics Daily, 2013. (8) Gholz, E. Rare Earth Elements and National Security; Council on Foreign Relationships, 2014. (9) Machacek, E.; Fold, N. Alternative value chains for rare earths: The Anglo-deposit developers. Resour. Policy 2014, 42, 53−64. (10) Golev, A.; Scott, M.; Erskine, P. D.; Ali, S. H.; Ballantyne, G. R. Rare earths supply chains: Current status, constraints and opportunities. Resour. Policy 2014, 41, 52−59. (11) Rademaker, J. H.; Kleijn, R.; Yang, Y. Recycling as a strategy against rare earth element criticality: A systemic evaluation of the potential yield of NdFeB magnet recycling. Environ. Sci. Technol. 2013, 47, 10129−10136. (12) Sprecher, B.; Kleijn, R.; Kramer, G. J. Recycling potential of neodymium: the case of computer hard disk drives. Environ. Sci. Technol. 2014, 48, 9506−9513. (13) Fiksel, J. Sustainability and resilience: toward a systems approach. Sustainability: Sci. Pract. Policy 2006, 2, 14−21. (14) Fiksel, J. Designing resilient, sustainable systems. Environ. Sci. Technol. 2003, 37, 5330−5339. (15) DeAngelis, D. L. Dynamics of Nutrient Cycling and Food Webs; Chapman & Hall: Boca Raton, FL, 1992; Vol. 9. (16) Pruyt, E. Small System Dynamic Models for Big Issues: Triple Jump towards Real-World Complexity; 1st ed.; TU Delft Library: Delft, 2013. (17) Brunner, P. H.; Rechberger, H. Practical Handbook of Material Flow Analysis; Lewis Publishers: Boca Raton, FL, 2003. (18) Bruneau, M.; Chang, S. E.; Eguchi, R. T.; Lee, G. C.; O’Rourke, T. D.; Reinhorn, A. M.; Shinozuka, M.; Tierney, K.; Wallace, W. A. Winterfeldt, von, D. A framework to quantitatively assess and enhance the seismic resilience of communities. Earthquake Spectra 2003, 19, 733−752. (19) Bruneau, M.; Reinhorn, A. Overview of the resilience concept. Proceedings of the 8th U.S. National Conference on Earthquake Engineering, San Francisco, April 18−22, 2006; Paper No. 2040. (20) Daigo, I.; Nakajima, K.; Fuse, M.; Yamasue, E.; Yagi, K. Sustainable materials management on the basis of the relationship between materials’ properties and human needs. Mater. Tech. (Les Ulis, Fr.) 2014, 102, 506. (21) Grimm, V.; Wissel, C. Babel, or the ecological stability discussions: an inventory and analysis of terminology and a guide for avoiding confusion. Oecologia 1997, 109 (3), 323−334. (22) Kondoh, M. Foraging adaptation and the relationship between food-web complexity and stability. Science 2003, 299 (5611), 1388− 1391. (23) Allison, G. The influence of species diversity and stress intensity on community resistance and resilience. Ecol. Monogr. 2004, 74, 117− 134. (24) Levin, S. A.; Carpenter, S. R.; Godfray, H. C. J.; Kinzig, A. P.; Loreau, M.; Losos, J. B.; Walker, B.; Wilcove, D. S. The Princeton Guide to Ecology; Princeton University Press: Princeton, NJ, 2009. (25) European Wind Energy Associaton. Wind EnergyThe Facts; 2009. (26) Polinder, H.; Ferreira, J. A.; Jensen, B. B.; Abrahamsen, A. B.; Atallah, K.; McMahon, R. A. Trends in Wind Turbine Generator Systems. IEEE J. Emerg. Sel. Topics Power Electron. 1, 174−185. (27) Lynas Corp, Siemens To Form JV For Magnet Production. Dow Jones International News, 2011. (28) Sprecher, B.; Xiao, Y.; Walton, A.; Speight, J.; Harris, R.; Kleijn, R.; Visser, G.; Kramer, G. J. Life cycle inventory of the production of rare earths and the subsequent production of NdFeB rare earth permanent magnets. Environ. Sci. Technol. 2014, 48, 3951−3958. (29) Du, X.; Graedel, T. E. Global in-use stocks of the rare earth elements: A first estimate. Environ. Sci. Technol. 2011, 45, 4096−4101. (30) Wardekker, J. A.; de Jong, A.; Knoop, J. M.; van der Sluijs, J. P. Operationalising a resilience approach to adapting an urban delta to
acknowledges but does not take into account the fact that nonlinearity plays an important role in complex systems; this framework incorporates nonlinearity through the explicit use of feedback loops. Furthermore, Greadel et al. consider environmental implications to be a dimension of criticality. It is an unfortunate fact that our interviewees indicated environmental considerations to be of little importance to their decision making process, and therefore is not explicitly included in our resilience framework. Despite the differences, there is a clear overlap between our framework and the frameworks proposed in the critical materials literature. The question is then: how do resilience and criticality relate to each other? We would go so far as to argue that one can define the criticality of a material in terms of how resilient its supply chain is. In a follow up paper, we will investigate the speed and magnitude of the system responses during the 2010 REE crisis, and provide indicators for the resilience mechanisms described in this work. Further development of supply chain resilience theory could greatly benefit from a body of mathematical work in theoretical ecology, that provides in-depth analysis of the causes and consequences of resistance, rapidity-like measures of recovery speed and flexibility, in complex ecosystems and food webs.15,21−24 It would be very interesting to see this framework applied to other supply chains than that of NdFeB magnets. We hope this paper will enable other researchers to look at leverage points, bottlenecks, and to develop policy options that take into account the full system surrounding the supply chains of critical materials.
■
ASSOCIATED CONTENT
* Supporting Information S
Interview list; cover letter; and semi-structured interview. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.5b00206.
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +31 (0)71 527 7461; e-mail: sprecher@cml. leidenuniv.nl. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We would like to thank Ibuki Komatsu for translating during the Japanese interviews, Van Gansewinkel Groep for their financial support and the anonymous reviewers for their detailed and constructive criticism. This research was carried out under project number M41.5.10408 in the framework of the Research Program of the Materials innovation institute (M2i).
■
REFERENCES
(1) Tukker, A. Rare earth elements supply restrictions: Market failures, not scarcity, hamper their current use in high-tech applications. Environ. Sci. Technol. 2014, 48, 9973−9974. (2) Report on critical raw materials for the EU; The Adhoc Working Group on Defining Critical Raw Materials, 2014. (3) Critical Materials Strategy; U.S. Department of Energy, 2011. (4) Seo, Y.; Morimoto, S. Comparison of dysprosium security strategies in Japan for 2010−2030. Resour. Policy 2014, 39, 15−20. 6749
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
Article
Environmental Science & Technology uncertain climate changes. Technol. Forecast. Social Change 2010, 77, 987−998. (31) Kingsnorth, D. Illegal Rare Earths Mining in China: A Threat to Long Term Planning & Sustainability. ERECON Final Conference, Milan, October 16th, 2014. (32) Bradsher, K. Main victims of mines run by gangsters are peasants.New York Times, 2010. (33) India backtracks on involving private miners in monazite. Mining Weekly, 2012. (34) Kato, Y.; Fujinaga, K.; Nakamura, K.; Takaya, Y.; Kitamura, K.; Ohta, J.; Toda, R.; Nakashima, T.; Iwamori, H. Deep-sea mud in the Pacific Ocean as a potential resource for rare-earth elements. Nat. Geosci. 2011, 4, 535−539. (35) Binnemans, K.; Jones, P. T.; Acker, K.; Blanpain, B.; Mishra, B.; Apelian, D. Rare-earth economics: The balance problem. JOM 2013, 65, 846−848. (36) Allwood, J. M.; Ashby, M. F.; Gutowski, T. G.; Worrell, E. Material efficiency: providing material services with less material production. Philos. Trans. R. Soc. A 2013, 371, 20120496−20120496. (37) U.S. Lobbying Group Warns Against Rare Earth Stockpile Plan. Rare Earth Investing News, 2013. (38) (Law Concerning the Japan Oil, Gas and Metals National Corporation). (39) Rare earth prices surge; funds pour in amid speculative concerns. Xinhua News Agency, 2011. (40) Smith, J. L. Resources policy. Resources Policy 2013, 38, 320− 331. (41) Mordike, B. L. Creep-resistant magnesium alloys. Mater. Sci. Eng.: A 2002, 324, 103−112. (42) Friedrich, H. E.; Mordike, B. L. Magnesium Technology; Springer: New York, 2006. (43) Bradsher, K. China Tries to Clean Up Toxic Legacy of Its Rare Earth Riches.New York Times, 2013. (44) Graedel, T. E.; Barr, R.; Chandler, C.; Chase, T.; Choi, J.; Christoffersen, L.; Friedlander, E.; Henly, C.; Jun, C.; Nassar, N. T.; et al. Methodology of metal criticality determination. Environ. Sci. Technol. 2012, 46, 1063−1070. (45) Nassar, N. T.; Du, X.; Graedel, T. E. Criticality of the rare earth elements. J. Ind. Ecol. 2015, DOI: 10.1111/jiec.12237.
6750
DOI: 10.1021/acs.est.5b00206 Environ. Sci. Technol. 2015, 49, 6740−6750
